helplog.25: rspectext, min_lenuserarea, out of space in image header

Frank Valdes wrote on Feb 22, 2005

NUMBER:	25
KEYWORDS:	rspectext, min_lenuserarea, out of space in image header
DATE:	Tue Feb 22 09:54:07 MST 2005
FROM:	valdes

Q:	I have a lot of spectra in simple column format, wavelength
	followed by flux. Many are not continuum normalized, and I would
	like to use IRAF for this task. As far as I can tell, I must first
	convert these images to an IRAF-readable format, but when I try to
	use rspectext for this purpose, retaining the wavelength
	information by using the nonlinear setting for dtype, I get the
	error message

	    ERROR on line 107: Out of space in image header (WAT2_774)

	I assume that I can set the relevant buffer length somehow, but I
	have not been able to find out where this setting is done from the
	documentation I have or from your FAQ's. Any suggestions would be
	appreciated!

A:	 You are correct that IRAF does not work directly off a text file
	such as you have.  It is a goal to do this someday but not much
	progress is being made on the goal right now.  So you would need to
	convert to the current format based on an image with the
	coordinates described in header keywords.

	The IRAF spectral formats are generally geared to using a
	functional representation of the wavelengths.  It does have a
	special case to allow listing the wavelengths but this has two
	limitations.  One is fundamental that the number of WAT keywords
	has to be less than 999.  The other is that the header space be
	allocated large enough to hold the number of keywords.  This latter
	consideration can be controled through an IRAF environment
	variable.  Look in your login.cl and you will see min_lenuserarea.
	This is the maximum number of characters preallocated by IRAF tasks
	whenever they read an image.  Since it looks like you are getting
	up to 774 cards you could uncomment the keyword and double the
	value.  Note that it is not a good idea to arbitrarily make this
	very large since in some cases, like combining many images, the
	memory allocated for the headers could cause programs to run out of
	memory.  But a factor of two increase would not be out of line.
	Let me know if this doesn't fix this particular error message.

	My other comment is that unless you suspect the wavelengths are not
	smoothly varying you might consider using the functional fit
	representation of the wavelengths by chosing "interp".  For many
	cases the difference between a fit and the explicit wavelengths is
	insignificant.  Often the way the wavelength were generated in data
	such as you have was done by fitting a function and then creating
	the format by evaluating the function at the pixels.

---

Last post on Feb 22, 2005